Abstract: Abstract: The planting of double cropping rice is a cropping system of planting and harvest twice a year in China. It is essential to get the planting area and spatial distribution of paddy rice at large scale for guiding rice production and regulating the regional balance of supply and demand. Currently, using remote sensing technology to monitor the rice planting in a wide range is becoming an increasingly important tool. However, the mixed pixel problem makes it imprecise to extract the double cropping rice paddy. In order to solve the mixed pixel problems on extracting planting area of paddy rice at large scale, a method was proposed based on similar index and linear spectral mixture model. The time-series of MODIS-EVI index was calculated by using multi-temporal MODIS09A1 data from 2010-4-15 to 2010-10-31 in Jiangxi province. The influence factors such as cloud were reduced by using Savizk-Golay filtering method. Combined with field works and HJ-1A CCD2 images, the rice field samples were identified according to the rice growth patterns. Then the standard double cropping rice EVI curve was extracted, and the similar index between each pixel's EVI value of MODIS images, and standard double cropping rice EVI curve was calculated. To construct the similarity index map of rice, the suspected pixels in double cropping rice areas were extracted, and each mutually independent spectrum was got based on minimum noise fraction separating principal component and noise. The pixel points which pixel purity index larger than 3.0 were selected by calculating the image pixel purity index and extracting the high purity pixel, and the results of N-dimensional divergence were analyzed using N-dimensional visualization tools. In order to test the accuracy of the extracting method, the HJ-1A CCD2 datum was used to carry on the spatial contrast verification, and the result showed that it was substantially coincide in spatial distribution. Compared with the Statistical Yearbook of Jiangxi Province in 2010, the extraction accuracy was 93%, the correlation of the regional statistical was satisfied with R2=0.9659. The study can provide a reference for the extraction of high precision rice information in the future.